An essential component of the new Guide for mechanistic–empirical design of new and rehabilitated pavement structures for the design of flexible pavement structures is the use of resilient modulus for base / subbase materials and subgrade soils. This study reports on resilient modulus (Mr) test results for unbound pavement materials that were obtained according to the American Association of State Highway and Transportation Officials (AASHTO) standard T307–99. Laboratory tests were performed on 36 representative aggregates from across Ontario and empirical relations between Mr and the bulk stress were investigated, as well as the sensitivity of Mr to moulding water content and gradation. This paper proposes to replace the nonlinear relation between resilient modulus and bulk stress with a linear relation between the two, taking into account the uncertainties that include the effect of varying water content through stochastic analysis. The effects of deviatoric stress on resilient modulus were found to be negligible for the granular aggregates that were tested. The use of a linear relation was computationally more efficient than the use of a nonlinear law; however, differences in strain predictions were observed. The findings from the finite element simulations were consistent to other studies that compared solutions using various constitutive models.